Increased levels of Body Mass Index (BMI) are associated with increased mortality and morbidity from cardiovascular disease, hypertension, diabetes and other disorders. The frequency of obesity and its associated health-related problems is increasing in the American population. We propose to identify DNA polymorphisms associated with increased BMI, with the goal of identifying genes involved in obesity. The application builds upon a two-stage genome scan for BMI performed in the NHLBI Family Heart Study (FHS). In the first, we examined 101 pedigrees with 1027 persons genotyped and found a LOD of 2.2 on chromosome 7. In stage 2 we examined 135 sibships, 380 persons, and found a LOD of 3.2 for the same locus. Compelling linkage was found in the combined study (LOD = 4.9, chr 7q31.3, 137cM). Identifying genes responsible for complex traits such as BMI has proven remarkably difficult. We propose a novel strategy which combines three cutting edge methods: (1) Regression Tree analyses to identify a homogenous subset of families with evidence for BMI linkage to 7q31.3; (2) DNA pooling of samples from linked versus unlinked families; and (3) quantitative PCR of DNA pools for very high-density SNP mapping. We believe that the combination of these methods will permit a cost effective approach for the identification of genetic polymorphisms in linkage disequilibrium with BMI, and has the potential to become a widely adopted method for gene localization of complex traits. Members of our investigative team have taken a lead role in designing the Regression Tree method for the identification of homogeneous subsets of linked families, based upon differences in anthropometry, lifestyle and physiologic factors. This strategy is designed to optimally divide the families into etiologically homogeneous subgroups by recursive partitioning, defined in terms of the strength of the linkage signal. The recursive partitioning technique dramatically increases power to detect linkage in a sample size such as that of the FHS. Methods for DNA pooling and allele frequency quantification by Mass Spectroscopy are equally fundamental to our application. Using DNA pools of 200 samples each, pools will be created from groups of obese and groups of non-obese individuals from families showing evidence for linkage to the 7q31.3 locus, and contrasting these with pools of obese and non-obese individuals in families that do not show evidence of linkage. Because a single PCR reaction represents the analysis of 200 study participants, fine mapping at a density of one SNP per 50Kb across a region of 20 to 40 Mb becomes readily achievable. We believe that this application has potential to develop methodologies for gene localization and identification for BMI and other complex traits.